Distilbert-base-financial-relation-extraction Open Source Model - Free for Financial Relation Detection and Classification

Distilbert Base Financial Relation Extraction

Developed by yseop

FReE is a model based on DistilBERT, specifically fine-tuned for financial relationship type detection and classification tasks.

Knowledge Graph

Transformers

#Financial Terminology Relationship Classification #DistilBERT Fine-tuning #High-precision Relationship Detection

Downloads 32

Release Time : 3/2/2022

Model Overview

This model is used to detect whether relationships exist between financial terms and classify them when relationships are present. It supports five relationship types: no relation, has, belongs to, is, and as.

Model Features

Specialized for Financial Domain

Optimized for financial terminology and relationships, suitable for financial text analysis.

Efficient Model Architecture

Based on DistilBERT, reducing computational resource requirements while maintaining performance.

Multi-source Data Training

Trained using financial data from various sources including Wikimedia, International Financial Reporting Standards, and Investopedia.

Model Capabilities

Financial Relationship Detection

Financial Relationship Classification

Text Classification

Use Cases

Financial Text Analysis

Financial Terminology Relationship Identification

Identify specific relationship types between financial terms

F1 score reaches 0.8243 (weighted average)

Financial Document Processing

Automatically process term relationships in financial documents

🚀 FReE (Financial Relation Extraction)

FReE is a DistilBERT base model. It's fine - tuned on a custom financial dataset for financial relation type detection and classification.

🚀 Quick Start

We present FReE, a DistilBERT base model fine - tuned on a custom financial dataset for financial relation type detection and classification.

✨ Features

Process

Detect the presence of a relationship between financial terms and qualify the relationship if it exists. Example use cases:

An A - B trust is a joint trust created by a married couple for the purpose of minimizing estate taxes. (Relationship exists, type: is)
There are no withdrawal penalties. (Relationship does not exist, type: x)

Data

The data consists of financial definitions collected from different sources (Wikimedia, IFRS, Investopedia) for financial indicators. Each definition has been split into sentences, and term relationships in a sentence have been extracted using the Stanford Open Information Extraction module. A typical row in the dataset consists of a definition sentence and its corresponding relationship label. The labels were restricted to the 5 most - widely identified relationships, namely: x (no relationship), has, is in, is and are.

Model

The model used is a standard DistilBERT - base transformer model from the Hugging Face library. See HUGGING FACE DistilBERT base model for more details about the model. In addition, the model has been pretrained to initialize weights that would otherwise be unused if loaded from an existing pretrained stock model.

Metrics

The evaluation metrics used are: Precision, Recall and F1 - score. The following is the classification report on the test set.

Property	Details
Model Type	A standard DistilBERT - base transformer model fine - tuned on a custom financial dataset
Training Data	Financial definitions from Wikimedia, IFRS, Investopedia, split into sentences with relationships extracted by Stanford Open Information Extraction

relation	precision	recall	f1 - score	support
has	0.7416	0.9674	0.8396	2362
is in	0.7813	0.7925	0.7869	2362
is	0.8650	0.6863	0.7653	2362
are	0.8365	0.8493	0.8429	2362
x	0.9515	0.8302	0.8867	2362

macro avg	0.8352	0.8251	0.8243	11810
weighted avg	0.8352	0.8251	0.8243	11810

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご